Max Robotics — Global Service Robot Directory + US Certification

About

Dehui Kong is a leading researcher in the emerging field of visual affordance recognition, a critical area bridging computer vision, robotics, and human-computer interaction. Her work focuses on enabling machines to understand not just what objects are, but what actions they permit—and crucially, whether those interactions are currently occurring. Kong’s major contributions include pioneering deep learning-based methods for affordance detection, most notably through her ADOSMNet architecture, which innovatively uses object shape masks to guide feature encoders for more precise detection. She further advanced the field with OASNet, a network that jointly models visual features and relational semantic embeddings to recognize an object’s affordance state—determining if an object is currently being interacted with. This work addresses a critical gap in traditional affordance learning, which could only identify potential functions. Her comprehensive 2023 survey on deep learning for visual affordance recognition, already garnering 17 citations, has become a foundational reference for researchers entering this domain. Kong’s research is directly applicable to enabling more intuitive and safe robotic manipulation, where understanding an object’s current state is as vital as knowing its potential uses.